BotFighters, Autonomous Space Combat Arena

Let’s do real-time, arcade-like. project-based learning proposal with autonomous agents at the center, and still incorporating simulation, APIs, and visualization.

BotFighters !

Inspiration in very old school games

Project Summary

You will develop a real-time 2D space combat game, inspired by XPilot.

• The game simulates physics-based movement (inertia, rotation, thrust), and pits autonomous agent-controlled ships against each other in a battle arena.
• The simulation exposes an API to control agents.
• Later phases involve writing bots that consume the API to compete, explore strategies, or apply, for instance, reinforcement learning.
• Don’t worry, we will change stuff as we go.

Project Phases

• Phase 1: Game Simulation Engine + API (Weeks 1–3)
• Phase 2: External Agent Development (Weeks 3–5)
• Phase 3: Enhancements & Research (Weeks 5–)

Phase 1: gamesim

• Game Simulation Engine + API (Weeks 1–3)
• Goal: Create the game world, ship dynamics, and expose APIs.

Phase 1 Components

• Physics engine: Simulate ship movement, rotation, thrust, collisions
• Game world: 2D arena with obstacles, walls, and pickups (ammo, health)
• Game objects: Ships, bullets, mines, shields, etc.
• Sensors: API exposes data like:
  • Radar scan (entities in radius)
  • Current velocity/angle
  • Health, ammo, cooldown timers

Phase 1 Components (2)

• Actuators via API:
  • rotate(angle), thrust(force), fire(), dropMine(), activateShield()
• Real-time or tick-based update loop
• REST API:
  • Agents send commands, receive sensor updates each tick
• Visualization:
  • Real-time 2D graphics using pygame or other packages
  • Show ships, effects, health bars, and others

Phase 2: agents

• External Agent Development (Weeks 3–5)
• Goal: Write agents that interact with the API to control ships.

Phase 2: agents

Bot Capabilities:

• Rule-based agents (if enemy is near → fire)
• some physics
• Pathfinding & obstacle avoidance
• Strategy switching (aggressive vs evasive)
• Teams: cooperative strategies (advanced)

Possibilities:

• Each student writes their own bot and enters a battle royale
• Scoreboards and tournaments via automated match runner

Phase 3:

• Enhancements & Research (Weeks 5–)
• Goal: Make the game deeper, smarter, or more beautiful.

Possible Extensions:

• Machine learning:
  • Train bots using RL (e.g., Q-learning, PPO)
• Complex game features:
  • Fog of war
  • Powerups, hazards (black holes, laser turrets)
• Network multiplayer mode
• Replay system for games + analytics
• Scientific twist: Use the game to study swarm behavior, emergent tactics

Suggested Tech Stack

• Python for game engine:
  • pygame real time looping
  • pymunk Pymunk is a 2D physics engine written in Python, binding Chipmunk C physics library
  • Arcade Modern alternative to pygame. Better suited for structured games with tile maps, platforms, and better performance.
  • FastAPI you know this one
  • turn-based or realtime (consider learning asyncio)
• REST API
• GitHub for code collaboration
• Optional: Web front-end

pygame

arcade

pymunk

Learning Objectives

• Code python within a team context
• GitHub workflows: branching, collaboration, PRs
• Real-time simulation & game loop structure
• REST/WebSocket API design
• Multi-agent systems and sensor/actuator patterns
• Game physics & visualization
• AI programming & strategy development
• Experimentation and data analysis (especially in ML or tournament phases)

Questions